3.3. Modularity
Landmarks in the subsets of the a-priori modularity hypothesis Subset 2: 8, 9, 10, 11. Using two blocks Partial least squares (PLS) Overall strength of association between blocks: yielded lowest RV coefficients; Permutation test against the null hypothesis of independence with 10000 rounds of randomization were also extremely significant (P <.0001) in dorsal skull landmark analyses but P-value: 0.0687 (in ventral view) with anatomical proximity and developmental origin employed in module partitioning, while correlation of PLS scores between blocks (0.99). Comparison with alternative partitions: using only contiguous partitions. All appropriate partitions were evaluated considering partitions possessing RV less than or equal to the a-priori hypothesis, (with minimal RV) this outcome confirms hypothesis 4 of the study. Principal Component Analysis: PCA: Covariance Matrix and Asymmetry component PC1= 88.38% PC2=7.48% variance, % predicted: 10.0770% Asymmetrical components demonstrated lower RV coefficients than the symmetrical components.
Table 8 Modularity hypotheses tested for dorsal and ventral skull views in the African Pangolin (P, tricuspis ) Modularity hypothesis among contiguous landmarks evaluated by full enumerations of partitions with the eleven (11) and nine (9) landmarks divided into 2 subsets. Partition with minimal RV minimum co-variation